Server crash

mitu · 09-17-2004, 07:14 AM

Hi,

I was wondering if anyone could please offer a bit of guidance with this? My server crashed last night, and the logs indicate this just before the crash:

Quote:

Sep 17 04:03:33 werewolf kernel: ------------[ cut here ]------------
Sep 17 04:03:33 werewolf kernel: kernel BUG at vmscan.c:684!
Sep 17 04:03:33 werewolf kernel: invalid operand: 0000
Sep 17 04:03:33 werewolf kernel: ipt_TOS iptable_mangle ip_conntrack_ftp ip_conntrack_irc ipt_LOG ipt_limit ipt_multiport ip$
Sep 17 04:03:33 werewolf kernel: CPU: 0
Sep 17 04:03:33 werewolf kernel: EIP: 0060:[<c013b040>] Not tainted
Sep 17 04:03:33 werewolf kernel: EFLAGS: 00010206

It's the first time that this has occurred. What would be your suggested method of action regarding this? Is this really a kernel bug, and if so, would a kernel upgrade/patch fix this? Or is this due to something else entirely?

The server is running RedHat 9.

Any help would be very much appreciated. Thanks in advance!

amosf · 09-17-2004, 09:23 AM

What kernel version is it.... I'd assume it's a rh kernel, not a generic, so no good reporting it to linux-kernel...

Can you reproduce the error, or was it a one off?

mitu · 09-18-2004, 04:42 AM

Hi Tim, thanks for your response! No, it was a one-off error, and has not occured since (though it has only been 24 hours). My kernel version is 2.4.20-31.9.

amosf · 09-18-2004, 05:56 AM

One off crashes are always a pain as it could be something that might never happen again and due to some wayward cosmic rays or power glitch perhaps

I know I went through this with my smoothwall box three days after getting it going. Spent days chatting about the crash on the smoothie list but got nowhere... Abd the machine has never done it since

Been up for months since then

Let us know if it gets more regular. You could maybe try an upgrade kernel, but I'd only do it if it crashed again.